Cursor
We track 5 benchmarks that Cursor made or helped make. Its models are on its lab page.
2 in use 0 with public code 3 with a leaderboard here
Benchmarks
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| CursorBench 4.0 Ambiguous multi-file coding tasks from real Cursor sessions in Cursor's production harness. Agentic graders score correctness, with cost reported. | Fixing real repositories | Sep 2026 |  60.4% | In use |
| CursorBench 3.1 Agents do ambiguous multi-file coding tasks from real Cursor sessions in Cursor's harness. Version 3.1 added codebase understanding, bug finding, planning and code review problems. Agentic graders score correctness. | Fixing real repositories | May 2026 | 72.9% | Fading |
| CursorBench 3.2 Agents do ambiguous multi-file coding tasks from real Cursor sessions in Cursor's harness. Version 3.2 added instruction following and advanced tool use problems. Agentic graders score correctness. | Fixing real repositories | Jul 2026 | 73.4% | In use |
| Cursor Bench (2025 internal version) Agents answer real agent requests from Cursor engineers in Cursor's harness. Graders compare against hand-curated solutions and check use of the codebase's existing abstractions. | Fixing real repositories | Oct 2025 | Fading | |
| CursorBench Agents do ambiguous multi-file coding tasks from real Cursor sessions in Cursor's harness. Version 3.0 covers edit, refactor and bugfix problems. Agentic graders score correctness, with cost reported. | Fixing real repositories | Mar 2026 | 60.4% | System cards only |